Prognostic factors influencing fresh and frozen IVF outcomes: an analysis of the UK national database

Abstract

National registry data provides a valuable resource to quantify factors which influence the outcomes of IVF treatments. Multilevel logistic regression analyses for live birth and multiple births given a live birth were undertaken using data from the UK Human Fertilization and Embryology Authority's registry of treatments conducted between 2000 and 2005. This study analysed 119,930 fresh and 19,918 frozen transfers from 85,349 patients in 84 centres. As well as quantifying the effects of a range of previously identified prognostic factors, the analyses showed that embryo cryopreservation reduced the live birth rate substantially (odds ratio (OR) 0.60, 95% confidence interval (CI) 0.57-0.64) compared with typical fresh cycles. Prognostic factors for transfers following cryopreservation were largely similar to those for fresh cycles, with some evidence that female age is less important, and there were small differences in the effects of the numbers of embryos transferred and previous IVF attempts that can be ascribed to selection effects. No factors were identified which specifically predicted twin outcomes; patients with a high twin risk were those with a high chance of treatment success. After allowing for all prognostic factors, there remained clinically important variability between centres (median OR 1.4) and between patients (median OR 1.8). National registry data provides a valuable resource which can be used to determine to what extent clinical and patient characteristics influence the outcomes of IVF treatments. We fitted complex statistical models for live birth and multiple births to data from the UK national registry of treatments conducted between 2000 and 2005. The analysis included 119,930 fresh and 19,918 frozen transfers from 85,349 patients in 84 centres. As well as quantifying the effects of a range of previously identified prognostic factors, we were able to estimate the magnitude of the losses due to embryo freezing and thawing. The effects of clinical and patient characteristics on the outcomes for frozen transfers were largely similar to those for fresh cycles. No factors were identified which specifically predicted twin outcomes; patients with a high twin risk were those with a high chance of success. After allowing for all the identified prognostic factors, there remained clinically important variability between centres and between patients which may suggest that the patients and clinics differ in their chances of success due to characteristics which are not currently being measured.